Audience Dialogue

Measuring musical tastes

Case study 5

For some radio stations, most of their audience research effort goes into measuring the musical preferences of their listeners. These radio stations are usually commercial ones, most of the time broadcasting music.

Musical preferences can change quickly, specially amoung young people, so a station still playing last year's popular songs (or even last month's) can rapidly lose its audience. For a commercial station, losing audience means losing money, so there's tremendous pressure to keep the music popular.

In Western countries, the normal way of doing this research is by telephone. These surveys are done regularly - every month, even every week. (The younger a station's listeners, the more quickly their preferences will change.)

A common method is for interviewers to ring up listeners at random and ask their opinions of 50 to 100 songs. Sometimes the interviewers refer to the songs by name. Sometimes they play a short "hook" to the listener over the phone. The "hook" is a fragment of the song - usually the most memorable part - and runs for 10 to 15 seconds.

If the respondent hasn't heard the song before, no question is asked. But respondents who haveheard the song are asked a question something like this:

"What's your opinion of that song? Is it...
- one of your favourites, or
- do you like it quite a lot, or
- do you like it a little, or
- don't you like it, or
- do you strongly dislike it, or
- did you once like it but are now tired of it?"

If a high percentage of listeners rate a song as one of their favourites, the program director will move the song onto "high rotation" (in other words, it will be broadcast more often). And if a song on "high rotation" begins to have a lot of listeners saying they are tired of it, it will be moved to a lower rotation, and less often played. Songs that receive a high "strong dislike" rating are dropped from the playlist in case listeners switch off the station when they hear these songs.

Do you sometimes wonder why stations playing popular songs often broadcast such a narrow range of songs? The answer lies partly in a flaw in the survey method. The problem is that, in music, listeners tend to like what they know, and know what they like. It's a vicious circle. If the survey results are strictly applied, the playlist steadily becomes smaller, and the average song receives a higher liking level.

For a while, this tactic often works well. The station's audience will rise, for six months or a year. But then, listeners begin to tire of the same old songs, and start tuning in to other stations - or even switching the radio off, if they find no other stations suitable.

But it doesn't always follow that if people like a song very much, they want to hear it very often. So the result of this programming policy is that songs which many people once liked very much have been thrashed so often on radio that they are now no longer played. A good example is "Stairway to Heaven," by Led Zeppelin.

Yet other songs, which listeners might have liked very much if only they'd heard them a few times, are also dropped quickly because they don't make it into "high rotation".

Another problem is that this type of survey approach leads stations to broadcast an ever-narrower range of music - so in some markets, many types of music are never played at all on radio - even though quite a lot of people like to listen to them.

After experimenting with possible solutions, we came up with a new approach to surveying musical tastes. The only difference was that instead of asking one question about each song, we now asked two:

1. "How often have you heard this song before? Never, maybe a few times, quite often, or very often?

2. "How often do you want to hear it again? Never, maybe a few times, quite often, or very often?"

All respondents were asked about all songs. This meant that the interviews took longer and therefore the survey cost more - but the results were more stable, and the surveys didn't need to be done so often (except for listeners aged under about 18).

Instead of producing a list of the percentage of listeners liking each song, we produced a graph, with each song represented by a dot. Songs were grouped into four categories, based on which quadrant on the graph they fell into:

1. Seldom heard before, not often wanted in future ("flops")

2. Seldom heard before, often wanted in future ("potential hits")

3. Often heard before, not often wanted in future ("burned out")

4. Often heard before, often wanted in future ("evergreens")

A big advantage of this survey approach over the common one described above is that it identifies potential hits.

If a station is doing regular surveys, it is a good idea to develop a panel of listeners - questioning the same sample of listeners over and over again. With this method, it's possible to identify a group of listeners whom we label the "forerunners" - people who are the first to like a song which eventually becomes a hit. These are not always the people who hear the song first - those are usually the ones who listen to the station the most; their views are often not representative of all listeners.

- Dennis List